On Measuring Faithfulness of Natural Language Explanations

Large language models (LLMs) can explain their own predictions, through post-hoc or Chain-of-Thought (CoT) explanations. However the LLM could make up reasonably sounding explanations that are unfaithful to its underlying reasoning. Recent work has designed tests that aim to judge the faithfulness of either post-hoc or CoT explanations. In this paper we argue that existing faithfulness tests are not actually measuring faithfulness in terms of the models' inner workings, but only evaluate their self-consistency on the output level. The aims of our work are two-fold. i) We aim to clarify the status of existing faithfulness tests in terms of model explainability, characterising them as self-consistency tests instead. This assessment we underline by constructing a Comparative Consistency Bank for self-consistency tests that for the first time compares existing tests on a common suite of 11 open-source LLMs and 5 datasets -- including ii) our own proposed self-consistency measure CC-SHAP. CC-SHAP is a new fine-grained measure (not test) of LLM self-consistency that compares a model's input contributions to answer prediction and generated explanation. With CC-SHAP, we aim to take a step further towards measuring faithfulness with a more interpretable and fine-grained method. Code available at \url{https://github.com/Heidelberg-NLP/CC-SHAP}

ViLMA: A Zero-Shot Benchmark for Linguistic and Temporal Grounding in Video-Language Models

With the ever-increasing popularity of pretrained Video-Language Models (VidLMs), there is a pressing need to develop robust evaluation methodologies that delve deeper into their visio-linguistic capabilities. To address this challenge, we present ViLMA (Video Language Model Assessment), a task-agnostic benchmark that places the assessment of fine-grained capabilities of these models on a firm footing. Task-based evaluations, while valuable, fail to capture the complexities and specific temporal aspects of moving images that VidLMs need to process. Through carefully curated counterfactuals, ViLMA offers a controlled evaluation suite that sheds light on the true potential of these models, as well as their performance gaps compared to human-level understanding. ViLMA also includes proficiency tests, which assess basic capabilities deemed essential to solving the main counterfactual tests. We show that current VidLMs' grounding abilities are no better than those of vision-language models which use static images. This is especially striking once the performance on proficiency tests is factored in. Our benchmark serves as a catalyst for future research on VidLMs, helping to highlight areas that still need to be explored.

MM-SHAP: A Performance-agnostic Metric for Measuring Multimodal Contributions in Vision and Language Models & Tasks

Vision and language models (VL) are known to exploit unrobust indicators in individual modalities (e.g., introduced by distributional biases), instead of focusing on relevant information in each modality. A small drop in accuracy obtained on a VL task with a unimodal model suggests that so-called unimodal collapse occurred. But how to quantify the amount of unimodal collapse reliably, at dataset and instance-level, to diagnose and combat unimodal collapse in a targeted way? We present MM-SHAP, a performance-agnostic multimodality score that quantifies the proportion by which a model uses individual modalities in multimodal tasks. MM-SHAP is based on Shapley values and will be applied in two ways: (1) to compare models for their degree of multimodality, and (2) to measure the contribution of individual modalities for a given task and dataset. Experiments with 6 VL models -- LXMERT, CLIP and four ALBEF variants -- on four VL tasks highlight that unimodal collapse can occur to different degrees and in different directions, contradicting the wide-spread assumption that unimodal collapse is one-sided. We recommend MM-SHAP for analysing multimodal tasks, to diagnose and guide progress towards multimodal integration. Code available at: https://github.com/Heidelberg-NLP/MM-SHAP

VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomena

We propose VALSE (Vision And Language Structured Evaluation), a novel benchmark designed for testing general-purpose pretrained vision and language (V&L) models for their visio-linguistic grounding capabilities on specific linguistic phenomena. VALSE offers a suite of six tests covering various linguistic constructs. Solving these requires models to ground linguistic phenomena in the visual modality, allowing more fine-grained evaluations than hitherto possible. We build VALSE using methods that support the construction of valid foils, and report results from evaluating five widely-used V&L models. Our experiments suggest that current models have considerable difficulty addressing most phenomena. Hence, we expect VALSE to serve as an important benchmark to measure future progress of pretrained V&L models from a linguistic perspective, complementing the canonical task-centred V&L evaluations.

MAGMA -- Multimodal Augmentation of Generative Models through Adapter-based Finetuning

Large-scale pretraining is fast becoming the norm in Vision-Language (VL) modeling. However, prevailing VL approaches are limited by the requirement for labeled data and the use of complex multi-step pretraining objectives. We present MAGMA - a simple method for augmenting generative language models with additional modalities using adapter-based finetuning. Building on Frozen, we train a series of VL models that autoregressively generate text from arbitrary combinations of visual and textual input. The pretraining is entirely end-to-end using a single language modeling objective, simplifying optimization compared to previous approaches. Importantly, the language model weights remain unchanged during training, allowing for transfer of encyclopedic knowledge and in-context learning abilities from language pretraining. MAGMA outperforms Frozen on open-ended generative tasks, achieving state of the art results on the OKVQA benchmark and competitive results on a range of other popular VL benchmarks, while pretraining on 0.2% of the number of samples used to train SimVLM.

What is Multimodality?

The last years have shown rapid developments in the field of multimodal machine learning, combining e.g., vision, text or speech. In this position paper we explain how the field uses outdated definitions of multimodality that prove unfit for the machine learning era. We propose a new task-relative definition of (multi)modality in the context of multimodal machine learning that focuses on representations and information that are relevant for a given machine learning task. With our new definition of multimodality we aim to provide a missing foundation for multimodal research, an important component of language grounding and a crucial milestone towards NLU.

Seeing past words: Testing the cross-modal capabilities of pretrained V&L models

We investigate the ability of general-purpose pretrained vision and language V&L models to perform reasoning in two tasks that require multimodal integration: (1) discriminating a correct image-sentence pair from an incorrect one, and (2) counting entities in an image. We evaluate three pretrained V&L models on these tasks: ViLBERT, ViLBERT 12-in-1 and LXMERT, in zero-shot and finetuned settings. Our results show that models solve task (1) very well, as expected, since all models use task (1) for pretraining. However, none of the pretrained V&L models are able to adequately solve task (2), our counting probe, and they cannot generalise to out-of-distribution quantities. Our investigations suggest that pretrained V&L representations are less successful than expected at integrating the two modalities. We propose a number of explanations for these findings: LXMERT's results on the image-sentence alignment task (and to a lesser extent those obtained by ViLBERT 12-in-1) indicate that the model may exhibit catastrophic forgetting. As for our results on the counting probe, we find evidence that all models are impacted by dataset bias, and also fail to individuate entities in the visual input.

AMR Similarity Metrics from Principles

Different metrics have been proposed to compare Abstract Meaning Representation (AMR) graphs. The canonical Smatch metric (Cai and Knight, 2013) aligns variables from one graph to another and compares the matching triples. The recently released SemBleu metric (Song and Gildea, 2019) is based on the machine-translation metric Bleu (Papineni et al., 2002), increasing computational efficiency by ablating a variable-alignment step and aiming at capturing more global graph properties. Our aims are threefold: i) we establish criteria that allow us to perform a principled comparison between metrics of symbolic meaning representations like AMR; ii) we undertake a thorough analysis of Smatch and SemBleu where we show that the latter exhibits some undesirable properties. E.g., it violates the identity of indiscernibles rule and introduces biases that are hard to control; iii) we propose a novel metric S2match that is more benevolent to only very slight meaning deviations and targets the fulfilment of all established criteria. We assess its suitability and show its advantages over Smatch and SemBleu.